Boreal fires burn into carbon-rich organic soils, thereby releasing large quantities of trace gases and aerosols that influence atmospheric composition and climate. To better understand the factors regulating boreal fire emissions, we developed a statistical model of carbon consumption by fire for Alaska with a spatial resolution of 450 m and a temporal resolution of 1 day. We used the model to estimate variability in carbon emissions between 2001 and 2012. Daily burned area was mapped using imagery from the Moderate Resolution Imaging Spectroradiometer combined with perimeters from the Alaska Large Fire Database. Carbon consumption was calibrated using available field measurements from black spruce forests in Alaska. We built two nonlinear multiplicative models to separately predict above- and belowground carbon consumption by fire in response to environmental variables including elevation, day of burning within the fire season, pre-fire tree cover and the differenced normalized burn ratio (dNBR). Higher belowground carbon consumption occurred later in the season and for mid-elevation forests. Topographic slope and aspect did not improve performance of the belowground carbon consumption model. Aboveground and belowground carbon consumption also increased as a function of tree cover and the dNBR, suggesting a causal link between the processes regulating these two components of carbon consumption. Between 2001 and 2012, the median carbon consumption was 2.54 kg C m-2. Burning in land-cover types other than black spruce was considerable and was associated with lower levels of carbon consumption than for pure black spruce stands. Carbon consumption originated primarily from the belowground fraction (median Combining double low line 2.32 kg C m-2 for all cover types and 2.67 kg C m-2 for pure black spruce stands). Total carbon emissions varied considerably from year to year, with the highest emissions occurring during 2004 (69 Tg C), 2005 (46 Tg C), 2009 (26 Tg C), and 2002 (17 Tg C) and a mean of 15 Tg C year-1 between 2001 and 2012. Mean uncertainty of carbon consumption for the domain, expressed as 1 standard deviation (SD), was 0.50 kg C m-2. Uncertainties in the multiplicative regression model used to estimate belowground consumption in black spruce stands and the land-cover classification were primary contributors to uncertainty estimates. Our analysis highlights the importance of accounting for the spatial heterogeneity of fuels and combustion when extrapolating emissions in space and time, and the need for of additional field campaigns to increase the density of observations as a function of tree cover and other environmental variables influencing consumption. The daily emissions time series from the Alaskan Fire Emissions Database (AKFED) presented here creates new opportunities to study environmental controls on daily fire dynamics, optimize boreal fire emissions in biogeochemical models, and quantify potential feedbacks from changing fire regimes.

Carbon emissions from boreal forest fires are projected to increase with continued warming and constitute a potentially significant positive feedback to climate change. The highest consistent combustion levels are reported in interior Alaska and can be highly variable depending on the consumption of soil organic matter. Here we present an approach for quantifying emissions within a fire perimeter using remote sensing of fire severity. Combustion from belowground and aboveground pools was quantified at 22 sites (17 black spruce and five white spruce-aspen) within the 2010 Gilles Creek burn in interior Alaska, constrained by data from eight unburned sites. We applied allometric equations and estimates of consumption to calculate carbon losses from aboveground vegetation. The position of adventitious spruce roots within the soil column, together with estimated prefire bulk density and carbon concentrations, was used to quantify belowground combustion. The differenced Normalized Burn Ratio (dNBR) exhibited a clear but nonlinear relationship with combustion that differed by forest type. We used a multiple regression model based on transformed dNBR and deciduous fraction to scale carbon emissions to the fire perimeter, and a Monte Carlo framework to assess uncertainty. Because of low-severity and unburned patches, mean combustion across the fire perimeter (1.98 ± 0.34 kg C m-2) was considerably less than within a defined core burn area (2.67 ± 0.40 kg C m-2) and the mean at field sites (2.88 ± 0.23 kg C m-2). These areas constitute a significant fraction of burn perimeters in Alaska but are generally not accounted for in regional-scale estimates. Although total combustion in black spruce was slightly lower than in white spruce-aspen forests, black spruce covered most of the fire perimeter (62%) and contributed the majority (67 ± 16%) of total emissions. Increases in spring albedo were found to be a viable alternative to dNBR for modeling emissions.

Relationships between boreal wildfire emissions and day-to-day variations in meteorological variables are complex and have important implications for the sensitivity of high-latitude ecosystems to climate change. We examined the influence of environmental conditions on boreal fire emissions and fire contributions to regional trace gas variability in interior Alaska during the summer of 2013 using two types of analysis. First, we quantified the degree to which meteorological and fire weather indices explained regional variability in fire activity using four different products, including active fires, fire radiative power, burned area, and carbon emissions. Second, we combined daily emissions from the Alaskan Fire Emissions Database (AKFED) with the coupled Polar Weather Research and Forecasting/Stochastic Time-Inverted Lagrangian Transport model to estimate fire contributions to trace gas concentration measurements at the Carbon in Arctic Reservoirs Vulnerability Experiment-NOAA Global Monitoring Division (CRV) tower in interior Alaska. Tower observations during two high fire periods were used to estimate CO and CH4 emission factors. We found that vapor pressure deficit and temperature had a level of performance similar to more complex fire weather indices. Emission factors derived from CRV tower measurements were 134 ± 25 g CO per kg of combusted biomass and 7.74 ± 1.06 g CH4 per kg of combusted biomass. Predicted daily CO mole fractions from AKFED emissions were moderately correlated with CRV observations (r = 0.68) and had a high bias. The modeling system developed here allows for attribution of emission factors to individual fires and has the potential to improve our understanding of regional CO, CH4, and CO2 budgets.

Black carbon (BC) aerosol emitted by boreal fires has the potential to accelerate losses of snow and ice in many areas of the Arctic, yet the importance of this source relative to fossil fuel BC emissions from lower latitudes remains uncertain. Here we present measurements of the isotopic composition of BC and organic carbon (OC) aerosols collected at two locations in interior Alaska during the summer of 2013, as part of NASA's Carbon in Arctic Reservoirs Vulnerability Experiment. We isolated BC from fine air particulate matter (PM2.5) and measured its radiocarbon (Δ14C) content with accelerator mass spectrometry. We show that fires were the dominant contributor to variability in carbonaceous aerosol mass in interior Alaska during the summer by comparing our measurements with satellite data, measurements from an aerosol network and predicted concentrations from a fire inventory coupled to an atmospheric transport model. The Δ14C of BC from boreal fires was 131 ± 52‰ in the year 2013 when the Δ14C of atmospheric CO2 was 23 ± 3‰, corresponding to a mean fuel age of 20 years. Fire-emitted OC had a similar Δ14C (99 ± 21‰) as BC, but during background (low fire) periods OC (45 to 51‰) was more positive than BC (-354 to -57‰). We also analyzed the carbon and nitrogen elemental and stable isotopic composition of the PM2.5. Fire-emitted aerosol had an elevated carbon to nitrogen (C/N) ratio (29 ± 2) and δ15N (16 ± 4‰). Aerosol Δ14C and δ13C measurements were consistent with a mean depth of burning in organic soil horizons of 20 cm (and a range of 8 to 47 cm). Our measurements of fire-emitted BC and PM2.5 composition constrain the end-member of boreal forest fire contributions to aerosol deposition in the Arctic and may ultimately reduce uncertainties related to the impact of a changing boreal fire regime on the climate system.